While doing some interactive data analysis (in a Jupyter notebook) I often use a ClusterManagers.ElasticManager
in the master process and then connect a couple of remote workers using Distributed.start_worker(...)
.
Sometimes these notebooks and its associated Julia kernel is running for days although I do not work with them anymore. I do not care so much about this one jupyter kernel, but
Is there a way to automatically disconnect idle worker processes after they are idle for say 10 minutes?